Research On Premium Customers Choosing And Optimized Power Purchasing And Selling Of The Electricity Retail Company

Under the background of a new round of power system reform in China,the selling side is gradually liberalized,and the electric retail companies have entered the market.From the external environment,the competitive pressure brought by the increasing number of power selling companies and the high operation cost under the current market rules have posed a great challenge to the operation of the electric retail companies.From the perspective of internal management,in the electricity market,the electric retail companies are also faced with the transaction risk brought by the uncertainty of the electricity price on the spot market,and the potential operational risks brought by the uneven qualification of the agents.Therefore,based on the analysis and excavation of the user side,premium customers' choosing model of the retail company and the operation mode centered on the purchase and sale of electricity are studied in this paper.Firstly,aiming at the question of clustering load curves consisting of high-dimensional data,a clustering method based on improved spectral multi-manifold clustering(SMMC)was presented,which included extraction of typical daily load curve,clustering of load curve and evaluation of clustering effect.In order to obtain similarity matrix of improved SMMC algorithm,canonical warping distance was used to measure curves' similarity and Gaussian kernel was used to calculate local similarity.After clustering,clustering validity indexes were adopted to evaluate the clustering results and algorithm performance from three aspects of clustering effect,algorithm stability and computing time.The new method's obvious superiority compared with the traditional clustering algorithm is verified by an example,which provides a foundation of clustering algorithm for customer's electric feature mining in later chapters.Secondly,the user's multi dimensional feature evaluation model is established,and the user's feature mining is carried out from 5 aspects of curve type,response ability,economic value,credit rating and predictability of electricity.Indexes' quantitative values are obtained by analytic hierarchy process(AHP)and criteria importance though intercrieria correlation(CRITIC).Based on this,an optimal combination model of premium users is established,and an example is given to verify that the model can effectively extract high-quality users,providing a load curve basis of the combined user group for the purchase and sale of electricity models in later chapters.Finally,the multi-objective optimization decision model of purchase and sale is set up.The electric retail company buys electricity in the medium and long-term bilateral market,day-ahead market,real time market.Meanwhile,company signs demand response contract with users to optimize the power consumption curve and reduce the power purchase ratio in the real time market.The profit maximization of the electric company and users' electrical satisfaction maximization are taken as the optimization goal.The genetic algorithm is used to solve the model and the optimal solution is selected by the fuzzy membership degree method.It is proved by an example that the model can reasonably configure the cost of electricity purchase for the electric company.